I wonder if this kind of experimental approach just leads to software slowly getting worse and worse over time, in many small steps. I feel like human response

I wonder if this kind of experimental approach just leads to software slowly getting worse and worse over time, in many small steps. I feel like human response to changes like this isn't linear: make a thing slightly worse, and most people won't be annoyed enough to do the research, find and change the setting. If it's not a huge step back they'll just live with it, and maybe complain to their IT-savy friend, who will just nod and say "I know". On the other hand, people doing the experiment will see that most of their test subjects don't switch away from the change - mission accomplished, full steam ahead!